Rainfall partitioning is a fundamental terrestrial ecohydrological process governed by canopy structure and rainfall characteristics. However, how event-scale rainfall partitioning in apple orchards responds to the combined effects of different growth stages and rainfall categories remains unclear. During four growing seasons (2013–2016), the 8- and 18-year-old apple orchards on the Loess Plateau were divided into growth stage I (rapid foliar growth) and stage II (rapid fruit enlargement and maturation) according to seasonal leaf area index (LAI) dynamics. Rainfall events were classified into three rainfall categories based on rainfall amount (RA), intensity (RI), and duration (RD) (category I: lowest mean RA, RD and RI; category II: moderate mean RA and RI and the highest mean RD; and category III: highest mean RA and RI, and moderate mean RD). Canopy interception (Ic), stemflow (SF) and throughfall (TF) were systematically quantified under the combined effects of growth stages and rainfall categories, and Boosted Regression Trees with partial dependence plots (PDPs) were applied to identify the dominant controlling factors. Results showed that growth stage effects on rainfall partitioning were limited (p > 0.05), whereas rainfall category was more strongly associated with variation in Ic and TF in both orchards, with Ic decreasing and TF increasing from category I to III. Under the combined effects of growth stages and rainfall categories, Ic and TF showed the same pattern at both stages. At stage I, Ic in the young and mature orchards decreased from 23.51% to 5.24% and from 26.60% to 6.49%, respectively, while TF increased from 74.91% to 92.36% and from 71.97% to 90.91%, respectively. Similar changes in rainfall partitioning components were observed at stage II, and the combined effects were consistently stronger in the mature orchard. SF remained a minor component and showed a weaker rainfall-category response than Ic and TF. RA remained the dominant predictor of rainfall partitioning, especially in the young orchard, whereas Ic and SF in the mature orchard under specific stage-category combinations were increasingly affected by RI, canopy structural variables and microclimatic variables. PDPs revealed threshold responses, with Ic stabilizing at RA of about 3.0 mm in the young orchard and 4.5 mm in the mature orchard under stage I and category I, while LAI around 2.1 m²m⁻² marked the turning point for Ic and TF under stage II and category II. These results indicate that seasonal canopy development mainly modified the magnitude of rainfall-category responses rather than independently determining rainfall partitioning. These findings improve understanding of event-scale rainfall redistribution among Ic, SF and TF under varying phenological and rainfall conditions and support phenology-based modelling and water management.